ConferenceCalls

Introduction

The focus of this repository is to take you through a series of text-mining methods in Finance with an ultimate objective of conducting sentiment analysis.

In particular, we'll be exploring these methods on a collection of earnings calls transcripts that were made available to us by Motley Fool, an online financial advisor. It is not known when the actual earnings announcements took place, but it is assumed that the calls occured either at the same time or right after the earnings announcements.

Essentially, the objective of an earnings call is to provide investors with information regarding a company's past performance and future plans/projects, as well as answering questions from the media and related personnel. With company executives discussing pertinent matters related to the company's financial performace and general outlook. On a more technical note, earnings calls play a role in updating investors beliefs towards a given company, therefore allowing them to make better informed investment decisions. For this very reason, earnings calls are particularly valuable for investors when relevant prior non-public information is discussed/revealed.

Example earnings call:

The field of text-mining is incredibly diverse and can be segmented into seven practice areas:

1. Search and information retrieval

2. Document clustering

3. Document classification

4. Web mining

5. Information extraction

6. Natural Language Processing (NLP)

7. Concept extraction

While all seven practice areas are very much in use in differing areas within economics and finance, a significant amount of research today is focused on document classification, information retrieval and NLP.

Sentiment analysis is considered to fall under the umbrella of document classification. In a nutshell, it seeks to determine general sentiments, opinions, and affective states of people reflected in a corpus of text.

Some well-known classification algorithms: